Contact In the news Acknowledgements References Links Downloads Protocols Brainstorm Meetings

October 21st

During the week:

I was busy with the estimation of parameters in the steady state with four points in the new model with degradation.

This is the new model :

e= alpha * T^m / (K^m+T^m)*(Ki^p+T^p)

so we have 5 parameters (alpha, m, K, p, Ki) to estimate but where as we have just four points from luciferase results we can just estimate four parameters.

First I started with estimating alpha,m,K,Ki but couldn't get good results; then I tried alpha,m,K,p and again no improvement. I found out that there is not much sensitivity to alpha but m,K,p and Ki are important so I put some value for alpha and tried to estimate the four rest. Again the best list square residual for s29 (the pink one) was about 10 and for s34 (violet one) was about 4. :(

I tried to increase the number of data points by interpolating from the Excel plots and used 6 points instead of 4 but again the optimum solution from the model was not successful to follow the experimental data. :(

This could be due to:

1) The model is not suitable.

2) Measurement errors.

3) Lack of enough data. (The interpolated points were not really belonged to the system)

in all of the cases I also tried the lsqcurvefit algorithm but it did not work. The error was:

Optimization terminated: first-order optimality less than OPTIONS.TolFun, and no negative/zero curvature detected in trust region model.

I couldn't solve this problem. :(